• Server parameter file : (SPFILE) See note 601157

• Automatic Undo Management : See note 600141

• Dynamic SGA : See note 617416

• Automatic PGA Memory Management : See note 619876

• If required, ASSM : See note 620803

1. SPFILE (from oss note 601157)
   a. Create SPFILE
   SQL>connect / as sysdba
   SQL>create spfile from pfile;
   b. renaming the old init.ora (for example):
   mv $ORACLE_HOME/dbs/init.ora $ORACLE_HOME/dbs/init.ora.SAVE
   c. SQL> connect/as sysdba
   SQL> shutdown
   SQL> startup
   SQL> show parameter spfile

2. Auto Undo management READ note 600141 below is an overview.
   Calculate the size of the Undo tablespace

   a. Step: Generate spool file (optional)
   Used to log the migration steps.
   SQL> spool undo_process.log
   b. Step: Detect the current management Undo mode
   SQL> show parameter undo
   SQL> select segment_name, tablespace_name, status from dba_rollback_segs;

   c. Step: Create a new PSAPUNDO Undo tablespace
   - manually or using SAPDBA 6.20 or BRSPACE 6.40.
   - Size: as large as possible, at least as big as PSAPROLL
   The size can be specified in K or M.
   - AUTOEXTEND: OFF (see recommendations)

   SQL> create undo tablespace PSAPUNDO datafile /oracle/SMP/sapdata2/undo_1/undo.data1′ size 1600M reuse
   autoextend off;
   d. Step: Assign an Undo tablespace
   init.ora: undo_tablespace = PSAPUNDO
   spfile : SQL> alter system set undo_tablespace = PSAPUNDO scope = spfile;
   e. Step: Define the Undo retention time
   init.ora: undo_retention = 43200 # 12h as start value
   spfile : SQL>alter system set undo_retention = 43200 scope = spfile;
   F. Step: Activate AUM
   init.ora: undo_management = AUTO
   spfile : SQL>alter system set undo_management=auto scope = spfile;
   G. Step: Start instance
   SQL> shutdown immediate
   SQL> startup
   h. Step: Check the current AUM settings
   SQL> show parameter undo
   SQL> select segment_name, tablespace_name, status from dba_rollback_segs;
   SQL> show parameter rollback_segments
   i. CAUTION:
   Provided step J below has not carried out yet, you can switch from the
   Automatic Undo Management to the Manual Undo Management at any time
   within a short time. The following steps are necessary for this:

   SQL> connect / as sysdba
   SQL> alter system set undo_management=manual scope = spfile;
   SQL> shutdown immediate
   SQL> startup

   j. Step: Delete the old rollback segments
   Except for the SYSTEM rollback segment, the old rollback segments are
   now deleted. They are no longer required.
   SQL> select segment_name, tablespace_name, status from dba_rollback_segs
   where status = ‘OFFLINE’ and segment_name not like ‘_SYSSMU%$’;
   SQL> DROP ROLLBACK SEGMENT PRS_;

   k. Check results:
   SQL> select segment_name, tablespace_name, status from dba_rollback_segs;
   Only UNDOs should appear;

   l. Step: Adjust the profile parameters
   spfile : SQL> alter system reset rollback_segments scope = spfile sid=’*';
   spfile : SQL> alter system reset transactions scope = spfile sid=’*';
   spfile : SQL> alter system reset transactions_per_rollback_segment scope = spfile sid=’*';
   spfile : SQL> alter system reset max_rollback_segments scope = spfile sid=’*';

   m. Step: Delete old rollback tablespace
   SQL> DROP TABLESPACE PSAPROLL including contents and datafiles;

3. Dynamic SGA
   a. SQL> alter system set statistics_level=’TYPICAL’;
   b. Setting the new parameters (if using SPFILE)
   SQL> connect / as sysdba
   SQL> REM Delete old SGA parameter
   SQL> alter system reset buffer_pool_keep scope = spfile sid=’*';
   SQL> alter system reset buffer_pool_recycle scope = spfile sid=’*';
   SQL> alter system reset db_block_buffers scope = spfile sid=’*';
   SQL> REM Set new SGA parameter
   SQL> alter system set sga_max_size=1531868916 scope = spfile;
   SQL> alter system set db_cache_size=476577792 scope = spfile;
   SQL> REM statistics_level = ALL or TYPICAL
   SQL> alter system set statistics_level=’TYPICAL’ scope = spfile;
   If required:
   SQL> alter system set db_keep_cache_size=X scope = spfile;
   SQL> alter system set db_recycle_cache_size=X scope = spfile;
   c. shutdown db / restart.

4. Auto PGA
   a. SQL> alter system set workarea_size_policy=AUTO scope = spfile;
   b. SQL> alter system set pga_aggregate_target=256M scope = spfile;

This script requires the SAP SDK, which is available from SAP.  If you can’t find it let me know.

So, this script will do the following:

• Attempt to connect to sap using sapinfo command check to see if host is already known to be down, or if recovered from being down

• send pages / email to defined list

• Tracks systems that are already down so as not to resend email / pages

And it may do more, or less.. I don’t know.  If you have suggestions or changes please email them to me at nick dot wilkens at gmail dot com.

#!/bin/sh
# N. Wilkens
# 8.26.04

# Location to sapinfo from sap SDK
SAPINFO=/usr/sapsdk/bin/sapinfo
#OVMSG=/opt/OV/bin/OpC/opcmsg

 

#SAP_HOST_FILE has the followinf format:
#<INSTANCE_NO> <SID> <CI_HOSTNAME> [OPTIONAL_APPSERVER HOSTNAMES.]

# EXAMPLE:

#00 PRD sapprod sapproda1 sapproda2 sapproda3
#01 CRP sapcrp

SAP_HOST_FILE=/usr/local/bin/SAP_HOSTS.PROD

 

MAILGRP=emailaddr

# function set_flag, creates a flag fil
function set_flag {
HOST=$1
STATUS=$2

case $STATUS in
DOWN)   touch /tmp/${HOST}.flag ;;
UP)     rm -f /tmp/${HOST}.flag ;;
*)      echo “Not supported”    ;;
esac
}

function send_alert {
HOST=$1
STATUS=$2

echo “${HOST} is ${STATUS}” |mailx -s “${HOST} is ${STATUS}” ${MAILGRP}
#${OVMSG}
}

# Check_Instance Function
function check_instance {
SYSNO=${1}; shift
SID=${1}; shift
HOST_LIST=${@}
for host in ${HOST_LIST}
do
# if the host is already know to be down, then skip.
if [ !${FLAGGED} ]
then
${SAPINFO} ashost=${host} sysnr=${SYSNO} >> /tmp/sapinfo.${host}

        # if the sapinfo command determines the host is not available, then
if [ $? -ne 0 ]
then
# IF the host flag file does not exist, then
if [ ! -e /tmp/${host}.flag ]
then
sleep 5
${SAPINFO} ashost=${host} sysnr=${SYSNO} >> /tmp/sapinfo.${host}
if [ $? -ne 0 ]
then
status=DOWN
set_flag $host $status
send_alert $host $status
#echo “${host} is down”
rm /tmp/sapinfo.${host}

                 #mv /tmp/sapinfo.${host} /tmp/sapinfo.${host}.$$  #useful for debugging problems
fi
#else
#   echo “$host still down”
fi
#else if the host is available..
else
if [ -e /tmp/${host}.flag ]
then
status=UP
set_flag $host $status
send_alert $host $status
#echo “${host} is up”
else
#echo “$host still up”
rm /tmp/sapinfo.${host}
fi
fi
else
echo “FLAGGED.. ”
fi
done
if [ -e dev_rfc ]
then
rm dev_rfc
fi
}

#usage: check_instance <SID_LIST> # as defined above
cat ${SAP_HOST_FILE} |while read HOST_LIST
do
check_instance $HOST_LIST &
done

SQL> select name from v$datafile;

NAME
——————————————————————————–
/oracle/ACM/oradata/ACM/system01.dbf
/oracle/ACM/oradata/ACM/undotbs01.dbf
/oracle/ACM/oradata/ACM/sysaux01.dbf
/oracle/NRXD/oradata/ACM/users01.dbf <—- Datafile to be moved / renamed..

Now, the files to be moved are known, the database can be shut down:

SQL> shutdown

The file(s) can be copied to their destination:

$ cp /oracle/NRXD/oradata/ACM/users01.dbf \
/oracle/ACM/oradata/ACM/users01.dbf
$ sqlplus “/ as sysdba”

SQL> startup exclusive mount
SQL> alter database rename file ‘/oracle/NRXD/oradata/ACM/users01.dbf’ to
‘/oracle/ACM/oradata/ACM/users01.dbf’;
SQL> shutdown
SQL> startup
SQL> select name from v$datafile;

NAME
——————————————————————————–
/oracle/ACM/oradata/ACM/system01.dbf
/oracle/ACM/oradata/ACM/undotbs01.dbf
/oracle/ACM/oradata/ACM/sysaux01.dbf
/oracle/ACM/oradata/ACM/users01.dbf

..

cd /opt

1. wget http://www.r-fx.ca/downloads/apf-current.tar.gz
2. wget http://www.mnxsolutions.com/scripts/apf.debian.init

3. tar -xzvf apf-current.tar.gz
4. cd apf-0.9.6-1/

5. ./install.sh
You will recieve back messages like below.. Notice the cp error.

Installing APF 0.9.6-1: cp: cannot create regular file `/etc/rc.d/init.d/apf’: No such file or directory
Completed.

Installation Details:
Install path: /etc/apf/
Config path: /etc/apf/conf.apf
Executable path: /usr/local/sbin/apf
AntiDos install path: /etc/apf/ad/
AntiDos config path: /etc/apf/ad/conf.antidos
DShield Client Parser: /etc/apf/extras/dshield/

Other Details:
Listening TCP ports: 21,25,53,465,587
Listening UDP ports: 53,1812,1813,48921
Note: These ports are not auto-configured; they are simply presented for information purposes. You must manually configure all port options.

6. mv /opt/apf.debian.init /etc/init.d/apf
7. chmod +x /etc/init.d/apf
8. update-rc.d apf defaults
9. Edit /etc/apf/conf.apf to appropriate settings you may include Listening TCP/UDP from above if approp.
10. /etc/init.d/apf start

11. Validate, change /etc/apf/conf.apf DEVEL_MODE=”0″
12. restart apf; /etc/init.d/apf restart

— Configure / Install BFD
.. cd /opt
1. wget http://www.r-fx.ca/downloads/bfd-current.tar.gz
2. tar -xzvf bfd-current.tar.gz
3. cd bfd-0.9/

4. ./install.sh
5. vi /usr/local/bfd/conf.bfd

Modify SLOG=”/var/log/secure” to SLOG=”/var/log/auth.log”

6. cd /usr/local/bfd/rules
7. mv sshd /tmp/bfd.old.rules.sshd # Just in case
8. wget http://www.mnxsolutions.com/scripts/bfd.rules.sshd -O sshd ; this works out of the box (or it did) for debian sarge.
9. chmod 600 sshd

For ubuntu my sshd rule ended up like:

REQ=”/usr/sbin/sshd”
if [ -f "$REQ" ]; then
LP=”/var/log/auth.log”
TLOG_TF=”sshd”
TRIG=”3″
TMP=”/usr/local/bfd/tmp”

## SSH
ARG_VAL1=`$TLOGP $LP $TLOG_TF.1 | grep sshd | grep -viw “error: BIND” | sed ’s/::ffff://’ | grep -iw “Failed password for invalid user” | awk ‘{print$13″:”$11}’ | grep -E ‘[0-9]+’ > $TMP/.sshd`
ARG_VAL=`cat $TMP/.sshd`
fi

10. edit /etc/ssh/sshd_config and add the following:

UseDNS no

11. /etc/init.d/ssh restart

To redirect a website from http -> https I utilized mod_rewrite. And created a .htaccess file with the following contents.

$ cat .htaccess
Options FollowSymLinks
RewriteEngine On
RewriteCond %{SERVER_PORT} !^443$
RewriteRule ^(.*)         https://your_server.com/$1

– I am partial to using LVM based devices for creating my virtual machines. Think lvm snapshots for backups.

A setup I am working on now consists of 4 systems with a dedicated LVM device, and an lvm based mount point for file based VMs.

# vgs
VG #PV #LV #SN Attr VSize VFree
xen1vg 1 6 0 wz–n- 1.96T 1.47T

# lvs
LV VG Attr LSize Origin Snap% Move Log Copy%
vm1rootlv xen1vg -wi-a- 80.00G
vm2rootlv xen1vg -wi-a- 80.00G
vm3rootlv xen1vg -wi-a- 80.00G
vm4rootlv xen1vg -wi-a- 80.00G
xen_master_img-lv xen1vg -wi-a- 80.00G
xenimages1lv xen1vg -wi-ao 100.00G

- If you decide to use a file based VM, but use a directory other than /var/lib/xen/images to store your image files, then selinux will get in the way. Use the following:
# semanage fcontext -a -t xen_image_t “/<new_directory>(/.*)?”
# restorecon -R /<new_directory>/

- If something is not working check /var/log/audit/audit.log and ‘dmesg’, disable selinux: ’setenforce 0′.

- If your system is installing or using a file based cdrom device, and you need to insert or change to another disk:
# xm block-list <vmname>
Vdev BE handle state evt-ch ring-ref BE-path
768 0 0 1 -1 -1 /local/domain/0/backend/vbd/14/768
5632 0 0 1 -1 -1 /local/domain/0/backend/vbd/14/5632

# xm block-detach <vmname> 5632 -f
# xm block-attach <vmname> file:/xen-images/nextdisk.iso hdc:cdrom r

Threads_created details the number of threads that have been created since the MySQL server started, and Connections is the total number of client connections to the MySQL server since startup. To work out the thread cache hit ratio, we use this calculation:

100 – ((Threads_created / Connections) * 100)
100 – ((10 / 78298) * 100) = ~99.987 Thread cache hit ratio

The ideal situation is to get Threads_created as close as possible to thread_cache_size – no new connections having to wait for new thread allocation – staying as close to around a 99% hit ratio as you can.

Of course, these hit ratios can be skewed on a system that has recently been restarted.